Large number of children admitted to the stabilization centre suffered from diarrhea and HIV. The cohort also had higher prevalence of edema at admission. These factors were found to independently increase their risk of mortality during subsequent treatment at the unit. However, diarrhea was associated with the highest risk of mortality adjusted for other factors. The mortality rate observed in this study was higher than that recommended by SPHERE (less than 10%) for inpatient management of SAM. It also fell short of WHO suggestions of less than 5%. Similarly, the mortality rate was higher than that in other studies done in Sub-Saharan Africa; the risk of mortality in NRUs for HIV-positive children in Sub-Saharan Africa was found to be 33.6% (range: 23.6%-38.4%). The first week of inpatient stay was the most critical to the survival of children; most deaths occurred during this period. Admission fever was not a reliable predictor of mortality in these children.
The age of the children in our study was comparable to the ages of children in studies done by Bachou et al (2006)  and Sunguya et al (2006)  Children in the study by Maitland K et al (2006) were older (median age of 25 months, IQR 16-46). The predominant (68.9%) form of malnutrition in our study was kwashiorkor. The study by Maitland K et al (2006), however, found a relatively lower prevalence (42%) of kwashiorkor.
In a study by Maitland et al (2006), the mortality rate was reduced from 30% to 19% following a stricter application of WHO therapeutic guidelines. However, the mortality rate observed in UTH had persistently been above 30% (ward audit) despite efforts to adhere to the WHO treatment recommendations. Similar high mortality rates were observed in inpatient units in Malawi, indicating a possibility of regional variation in case presentation and response to treatment. For a 59-bed capacity ward, the UTH inpatient unit has more children than it can accommodate year around. The congestion of the ward, with more than one child per cot, made management of the children difficult. The presence of children's caretakers and accompanying siblings has made management even more challenging. This has negatively impacted on the quality of service. The impact of such operational conditions on outcome of children treated in inpatient setups has been shown in the paper by Heikens et al, (2008).
The HIV prevalence found among children in our study is higher than the 29.2% prevalence found in a recent meta-analysis. Moreover, both in our study and that of Fergusson, P. & Tompkins, A., (2009) it was found that HIV-positive children had a higher risk of death than HIV-negative children.
The high prevalence and observed effect of diarrhea on the outcomes of the children calls for strengthened community-level interventions, targeted towards prevention and treatment of diarrhea. Simple and effective interventions such as hand-washing, zinc supplementation, oral rehydration solutions, and water and sanitation interventions at the community level deserve a critical look. This is more relevant taking into consideration the poor sanitary condition of the areas the admitted children came from.
Since the advent of Community-Based Therapeutic Care (CTC), the case fatality of children with SAM without complications has been reduced to less than 5% [5, 26]. Accordingly more efforts need to be made in Lusaka to identify children suffering from SAM at an earlier stage by strengthening active case finding at the community level. In addition, the value of a supplementary feeding program, which is lacking in the current CTC program in Lusaka, needs to be looked into in order to prevent fast deterioration of children into severe malnutrition during peak malnutrition periods. Related to the documented high prevalence of kwashiorkor, the current practice of providing high doses of vitamin A as part of the inpatient management is questionable in the face of evidence demonstrating increased risk of mortality in children with edematous forms of malnutrition receiving high doses of vitamin A .
Limitations and strength
The fact that the HIV test results presented in this paper were not confirmed by PCR, while over half of the children included in the study were less than 18 months old is a serious shortcoming of our study. Lack of key socio-demographic data might have limited our understanding of the impact such factors play on the outcome of children. Exclusion of children admitted over the weekend might have introduced a selection bias. From observation children admitted over weekends tended to be more likely to be severely ill.
This study presents findings of children with complicated SAM that receive treatment in a context where ward congestion, staff turnover and fatigue, and limited diagnostic ability exist. In addition, children came from an environment where prevalence of infectious diseases such as diarrhea and HIV are high. The operational research nature of our study, representing a real life scenario, makes it representative of most conditions in developing countries. A renewed effort to better understand the appropriate management children with complicated SAM that present with diarrhea exists, and we believe our study will add an impetus to this effort.